Department of Political Science

CENTRAL EUROPEAN UNIVERSITY

THE RELEVANCE OF MICRO SOCIAL CONTEXTS FOR

INDIVIDUAL POLITICAL ENGAGEMENT: A COMPARATIVE

ANALYSIS

By

Oana Lup

A Doctoral Dissertation

Submitted to the Central European University in partial fulfillment of the

requirements for the Degree of Doctor of Philosophy

August, 2011

Supervisory committee:

Gábor Tóka, Central European University

Paul Allen Beck, Ohio State University

Tamás Rudas, Central European University& ELTE

I hereby declare that this work contains no materials accepted for any other

degrees in any other institutions. This thesis contains no materials previously

written and/or published by another person unless otherwise noted.

Oana Lup

August, 2011

i

Abstract

In my dissertation I explore the role played by the micro social contexts in which ordinary

citizens are embedded on their political engagement. I start my examination with the assumption

that participation in informal political conversations promotes individuals‟ political participation

and contribute to an increase in their level of political cognition. I analyze the effects of everyday

political discussion in two types of micro social settings – close, intimate and generic ones.

Results indicate that political discussion in either type of social setting has both direct and

indirect effects on voting and political knowledge. Most of the direct effects are moderated by

macro contextual factors. Specifically, frequency of political talk in intimate settings appears to

advance individual political knowledge in those countries characterized by extensive political talk

in the micro social settings. Political agreement with close, intimate peers stimulates electoral

participation, especially in newly democratized countries. Frequent political talk within generic

social settings, on the other hand, generates an increase in political knowledge in less developed

countries. With regard to the indirect effects, my results indicate that political discussion

stimulates individuals‟ attention to political news in media and general political interest. Political

interest and media attentiveness are, in turn, the most significant antecedents of voting and

political knowledge.

I continue with an empirical examination of the hypothesis that political discussion is a

significant antecedent of individual political engagement and test different directions of

influences between the two. Results indicate that the assumption present in the previous literature

might not be tenable in its original form. In the context of Japan, the classical direction of effects

seems to be reversed; more politically knowledgeable and opinionated people are more likely to

ii

engage in political conversations. On the other hand, there is evidence that informal political talk

influences individuals‟ levels of political opinionation and knowledge indirectly. More frequent

political discussion and membership in loosely connected networks of political conversation

strongly affects individuals‟ interest in politics, which, in turn, influences their level of political

knowledge. For the relationship between political discussion and participation, my results

confirm the expectations formulated by the previous literature; frequent political conversation

stimulates participation in political and civic activities in Hungary.

Having established that in some contexts and under specific circumstances micro social

settings play an important role in politics, I examine the supply side of those politically relevant

features of social networks, in countries dissimilar in their democratic experiences. The results

indicate that people from countries with a more recent democratic experience are, in general, part

of smaller, less diverse, less politicized and more politically homogeneous social networks

compared to their counterparts from older democracies. Within the group of newly democratized

countries, people who were socialized in a democratic period are similar to young generations

from consolidated democracies in their patterns of social connectivity. However, in the new

democracies, those generations who were socialized in democracy tend to participate less in

political conversations compared to those who became adults under authoritarian regimes.

iii

Acknowledgments

I want to thank those who lent their scholarly expertise and support at various stages of

writing my PhD dissertation. My supervisor, Gabor Toka, deserves great appreciation for his

constant and extremely valuable guidance from the early stages of my project. He taught me

many things from writing a proposal to dealing with various statistical methods, but, most

important, he taught me what rigorous research means. I cannot thank him enough for all the time

he spent and all the interest he put in my project.

Paul Allen Beck became involved later in supervising my PhD project and, from the very

beginning, we shared an interest for the study of social influences in politics. I would like to

thank him firstly for accepting to mentor me during my visit at Ohio State University and for

being always available to answer my questions and discuss my research problems. I am especially

appreciative of his constructive criticism and scholarly rigor. Also, I want to thank both him and

Richard Gunther for granting me access to the CNEP data that I used in my dissertation and for

assisting me in using it.

Perhaps one of the most challenging collaboration was the one with Tamas Rudas. The

best that can happen to a student is to be exposed to a different framework of approaching a

problem and this describes my experience with Tamas Rudas‟ way of looking at my project. I am

grateful to him for teaching me the values of good research and the importance of approaching

problems in an inquisitive manner. I hope that my best way of „repaying‟ him would be by

passing the same standards to my students, in the future.

I want to express my gratitude to Ken‟ichi Ikeda, for his numerous interventions in my

project. He gave me access to the Japanese data sets that I employed in my analysis, read my

iv

chapter on Japan and always gave timely and constructive feedback on my research. His scholarly

passion for the study of social networks is an example for me.

I benefitted a lot from my discussions with Robert Tardos. Not only did he help with

compiling some of the Hungarian data that I used in my dissertation but he also gave extremely

valuable insights into networks dynamic in the Hungarian context.

I learned a lot about how to use statistical methods from Levente Littvay. He was always

available when I needed his help on nitty-gritty problems with my statistical analysis. Also,

special thanks to Thomas Rudolph who always gave timely answers to my questions on

multilevel analysis.

During my visit at Ohio State University I had the opportunity to meet two very important

scholars, whose research is particularly relevant for me. I want to thank Erik Nisbet and Wiliam

„Chip‟ Eveland for advising me on my project. Wiliam „Chip‟ Eveland read an important part of

this dissertation and gave constructive and useful feedback. I owe special thanks to Philipp Rehm,

whose apparently small comment contributed to an important turn in my paper.

In the process of writing my dissertation I benefited from grants and fellowships offered

within EU framework programs. I participated in the Workshop „connecting Excellence on

European Governance‟ Connex „Civil Society Involvement in European Governance‟, hosted by

the Centre for European Social Research MZES, in Mannheim. I want to thank especially Jan van

Deth, Sonja Zmerli and Sigrid Roßteutscher for their helpful comments in the early stages of my

dissertation writing. Also, Jan van Deth gave me access to the CID data that I used in my

dissertation and advised me on various issues related to these data sets. I also benefited from the

visits at the European Centre for Analysis in the Social Sciences (ECASS) at the Institute for

Social and Economic Research, University of Essex, supported by the Access to Research

Infrastructures action under the EU Improving Human Potential Programme, and at the European

v

Data Laboratory for Comparative Social Research (EUROLAB) supported by the European

Community under the „Structuring the European Research Area‟ specific programme, Research

Infrastructures Action of the 6th Framework Programme. I owe special thanks to Ingvill

Mochmann for her assistance during my stay at the EUROLAB.

The whole process of thinking over and writing this dissertation was a lengthy one. Many

people helped with comments in formal and less formal settings. I owe special thanks to my

colleagues from the Political Behavior Research Group (PolBeRG), and to Balazs Vedres, Zsolt

Enyedi and Marina Popescu. Very often, casual conversations were extremely useful in sparking

arguments and helping me to reflect on the foundations of this project. Daniela Lup and Peter

Maas were my companions in this enterprise.

Last and for sure not least I owe many thanks to my academic writing instructors, John

Harbord and Thomas Rooney; I learned a lot from them. Oana Pop‟s help is greatly appreciated;

she was extremely patient and supportive in her help with editing this text.

Finally, this dissertation is dedicated to my parents who always trusted me.

vi

Table of contents

Introduction ...................................................................................................................................... 1

Chapter 1: Social networks, political discussion and engagement: the state of the art .................... 9

1.1. Main concepts ........................................................................................................................ 9

1.2. State of the art ...................................................................................................................... 16

1.3. Research questions, design and data ................................................................................... 26

Chapter 2: Micro social embeddedness, electoral behavior and political knowledge: a

comparative analysis ...................................................................................................................... 29

2.1. Micro social contexts, political behavior and knowledge: a review ................................... 31

2.1.1. Micro social contexts and individual turnout ............................................................... 32

2.1.2. Micro social contexts and political knowledge ............................................................ 39

2.2. Hypotheses .......................................................................................................................... 42

2.3. Data, measures and research design .................................................................................... 49

2.3.1. Data .............................................................................................................................. 50

2.3.2. Measures ....................................................................................................................... 51

2.3.3. Research design ............................................................................................................ 57

2.4. Results ................................................................................................................................. 60

2.4.1. Intimate social networks, voting and knowledge ......................................................... 60

2.4.1.1. Intimate social networks and voting ...................................................................... 61

2.4.1.2. Intimate social networks and political knowledge ................................................ 68

2.4.1.3. Intimate social networks and political interest ...................................................... 76

2.4.1.4. Intimate social networks and media attentiveness ................................................ 80

Chapter 3: Generic social settings, voting and political knowledge: a comparative analysis ........ 90

3.1. Generic social networks and individual turnout .................................................................. 90

3.2. Generic social networks and political knowledge ............................................................... 95

3.3. Generic social networks and political interest ................................................................... 101

3.4. Generic social networks and media attentiveness ............................................................. 104

3.5. Conclusions and limitations .............................................................................................. 109

Chapter 4: Political communication and cognitive involvement with politics: examining

directions of effects in Japan ........................................................................................................ 115

4.1. Direction of effects between political communication and cognitive involvement with

politics: the state of the art ....................................................................................................... 121

4.2. Research design, data and measures .................................................................................. 123

vii

4.2.1. Data ............................................................................................................................ 128

4.2.2. Measures ..................................................................................................................... 132

4.3. Political communication and cognitive involvement with politics: a non-recursive model

design ........................................................................................................................................ 138

4.4. Political communication and cognitive involvement with politics: a recursive model design

.................................................................................................................................................. 150

4.4.1. Casual political conversations and political opinionation: results ............................. 155

4.4.2. Casual political conversation and knowledge: results ................................................ 160

4.4.3. Political discussion and interest: results ..................................................................... 163

4.5. Conclusions and limitations .............................................................................................. 166

Chapter 5: Social networks and political participation: examining directions of effects in Hungary

...................................................................................................................................................... 172

5.1. Research design, data and measures .................................................................................. 174

5.1.1. Data ............................................................................................................................ 175

5.1.2. Measures ..................................................................................................................... 176

5.2. Interpersonal communication and political participation: a non-recursive research design

.................................................................................................................................................. 179

5.3. Interpersonal communication and political participation: a recursive research design ..... 183

5.4. Conclusions and discussion ............................................................................................... 192

Chapter 6: (Small) Talk that Matters: a comparative examination of social networks and political

communication ............................................................................................................................. 197

6.1. Determinants of social networks and political discussion: state of the art ........................ 200

6.2. Hypotheses, research design, and data .............................................................................. 205

6.3. Personal and political discussion networks: a comparative assessment ............................ 210

6.3.1. Size and diversity of social networks ......................................................................... 210

6.3.2. Politicization of networks ........................................................................................... 222

6.3.3. Political agreement ..................................................................................................... 228

6.3.4. Generational differences in social connectivity and political discussion patterns ..... 234

6.4. Conclusions ....................................................................................................................... 244

Conclusions .................................................................................................................................. 247

List of References ......................................................................................................................... 253

Appendices ................................................................................................................................... 264

viii

List of Tables and Figures

Table 1: Micro- and macro-level determinants of individual turnout ............................................ 67 Table 2: Micro- and macro-level determinants of political knowledge ......................................... 73 Table 3: Micro- and macro-level determinants of political interest ............................................... 79

Table 4: Micro- and macro-level determinants of media attentiveness ......................................... 84 Table 5: Micro- and macro-level determinants of individual turnout ............................................ 94 Table 6: Micro- and macro-level determinants of individual political knowledge ........................ 98 Table 7: Micro- and macro-level determinants of individual political interest ............................ 103

Table 8: Micro- and macro-level determinants of media attentiveness ....................................... 106 Table 9: Effects of political communication on opinionation, OLS estimates with robust SE .... 141 Table 10: Effects of political communication on knowledge ....................................................... 145

Table 11: Effects of political communication on interest ............................................................ 148 Table 12: Fit statistics for models of effects between political communication and opinionation

...................................................................................................................................................... 157 Table 13: Estimates of effects between political communication and opinionation .................... 159

Table 14: Fit statistics for models of effects between political communication and knowledge . 161 Table 15: Estimates of synchronous effects of political knowledge on communication ............. 162 Table 16: Fit statistics for models of effects between political communication and interest ...... 164

Table 17: Estimates of reciprocal effects between political communication and interest ........... 165 Table 18: Effects of political communication on political participation and interest, OLS

estimates with robust standard errors ........................................................................................... 181

Table 19: Fit statistics for models of effects between political discussion and participation ...... 187

Table 20: Estimates of reciprocal effects between political discussion and participation ........... 189 Table 21: Fit statistics for models of effects between social networks and political participation

...................................................................................................................................................... 190 Table 22: Estimates of reciprocal effects between social networks and political participation ... 191 Table 23: Average number of three types of friends (ISSP 1986) ............................................... 212

Table 24: Average number of three types of friends (ISSP 2001) ............................................... 213 Table 25: Test of significance of countries‟ differential effect on number of close friends ........ 214

Table 26: Measures of social networks (CID) .............................................................................. 217 Table 27: Test of significance of countries‟ differential effect on measures of social networks . 221 Table 28: Average measures of political talk in various social settings (standard errors in

parentheses) .................................................................................................................................. 225

Table 29: Test of significance of countries‟ differential effects on frequency of political talk ... 227 Table 30: Political agreement within intimate micro social settings ............................................ 230 Table 31: Level of political agreement within various generic groups ........................................ 233

Table 32: Determinants of frequency of general political discussion; OLS estimates ................ 237 Table 33: Determinants of general political discussion; OLS estimates ...................................... 239 Table 34: Determinants of membership to a group of friends; logistic regression estimates ...... 243

ix

Figure 1: Cross-lagged reciprocal effects between political discussion and cognitive involvement

...................................................................................................................................................... 152 Figure 2: Cross-lagged unidirectional effects from political discussion to cognitive involvement

...................................................................................................................................................... 152 Figure 3: Cross-lagged unidirectional effects from cognitive involvement to political discussion

...................................................................................................................................................... 153 Figure 4: Synchronous reciprocal effects between political discussion and cognitive involvement

...................................................................................................................................................... 154

Figure 5: Synchronous unidirectional effects from political discussion to cognitive involvement

...................................................................................................................................................... 154 Figure 6: Synchronous unidirectional effects from cognitive involvement to political discussion

...................................................................................................................................................... 155 Figure 7: Cross-lagged reciprocal effects between political discussion and participation ........... 184

Figure 8: Cross-lagged, unidirectional effects from political discussion to participation ............ 184 Figure 9: Cross-lagged, unidirectional effects from political participation to discussion ............ 184

Figure 10: Synchronous, reciprocal effects between political discussion and participation ........ 185 Figure 11: Synchronous, unidirectional effects from political discussion to participation .......... 185

Figure 12: Synchronous, unidirectional effects from political participation to discussion ........ 186

1

Introduction

Ordinary citizens‟ participation in political discussions has been seen as a desirable

feature of democracies ever since the heydays of the Athenian demos. More recently, advocates

of deliberative democracy have suggested that everyday political talks function as fora in which

people learn about politics, form and refine their political preferences, and prepare to express

them through political actions. In short, the give and take of everyday political conversations is

believed to lead to an increase in levels of political interest, knowledge and participation. A less

optimistic view about informal political conversation though, is that this is a mere form of social

entertainment and sometimes might hinder political knowledge and lead to political intolerance.

This happens when political discussion exposes people to one-sided views and insulates them in

politically homogeneous micro social settings, in which a dialogue between holders of divergent

political views becomes impossible.

In the previous literature, the empirical test of the statements that describe potential

benefits of informal political conversation revealed inconsistent findings, tradeoffs between

equally desirable political outcomes that are related to political talk and, for most of the cases,

several limitations imposed by available data. One of the most challenging conclusions was

reached by Mutz‟s study, which suggested that the type of political communication that advances

political participation – one that exposes people to similar political views – is incompatible with

the one that promotes political tolerance – political conversation that exposes people to dissimilar

views (Mutz 2006). This finding indicates that informal political discussion is not a panacea for

alleviating all democratic dysfunctions. On the contrary, it invites a more nuanced approach in

2

studying the role of political communication for individual political engagement. However,

Mutz‟s results are not conclusive. Other scholars found evidence that exposure to political

disagreement does not hinder political participation (Horan 1971, Nir 2005) and might even

promote it, through an increase in political knowledge (Leighley 1990, McLeod et al. 1999,

Scheufele et al. 2003, Scheufele et al. 2004, Kwak et al. 2005). Notwithstanding the

inconsistencies of their results, most of these studies cannot say much with regard to the direction

of the influences flowing between political communication and engagement. Although most of

them found a significant relationship between some features of informal political conversation

and political engagement, they were unable to establish whether political talk drives political

engagement or, on the contrary, political engagement leads to an increase in levels of informal

political conversation.

The study of the relationships between social networks and political engagement is worth

pursuing for both theoretical and practical reasons. From a theoretical point of view, it is

important to clarify the role of interpersonal communication among other classic predictors of

individual political engagement. Previous research documented the existence of strong

relationships between the micro social settings in which individuals are embedded and their levels

of political participation and cognition. Therefore, empirical models that fail to consider the

social nature of individual political engagement (and fail to include features of political

discussion in social networks among other explanatory variables) are underspecified and lead to

biased estimates and erroneous conclusions. To give an example, some studies showed that, when

social networks are included in models that predict political behavior and knowledge, the effects

of classical predictors, such as education and income drop or diminish. This suggests that the

mechanism through which these demographics operate is by placing people in more politicized

settings, for instance (Abrams et al. 2010). Equally important, thus, the inclusion of sociological

3

factors into models of political engagement should be accompanied by an explanation of the

mechanisms that make political communication a significant antecedent of political participation

and knowledge. Besides these technical concerns about omitting social context from empirical

models of individual political behavior, there are equally important practical ones. These

theoretical aspects have practical consequences. Understanding the circumstances and the

mechanisms that make participation in political conversations a significant antecedent of political

engagement can inform policies designed to create a more politically educated and participative

citizenry. Such measures should be preceded by a comprehensive examination of the effects that

various features of political communication might have on forms of political engagement. As

Mutz pointed out, equally valuable political outcomes, such as participation and political

tolerance, for instance, appear to be served by different informal discursive practices. If this is

indeed the case, social scientists and public policy pundits should find effective ways to equip

with their citizens so that they get along with exposure to political disagreement in their social

networks while keeping social cohesion and harmony in their social interactions.

There is abundant evidence that the way individuals understand various facts and behave

in different realms of their lives is both a matter of their individual characteristics and of the

social contexts to which they belong. People do not make decisions – be they political or not – in

isolation but as they find themselves embedded in diverse social networks they are susceptible to

employ and adopt the ideas and norms that flow through those groups. In short, people are

connected and thus, the understanding of their attitudes and behaviors should take into account

the structure of the linkages among them and what flows within those groups, i.e. ideas, social

norms (Christakis and Fowler 2009).

My dissertation contributes to existing research on these topics in the following ways.

First, my study enlarges the scope of previous analysis by examining the relationship between

4

interpersonal communication and political engagement across countries that differ in their

histories and political cultures. This comparative assessment makes it possible to answer the

question whether these relationships are general or context dependent. Previous studies were

conducted in a limited number of countries, mostly the US and more recent in the UK, Germany,

Spain and Japan, and therefore there is a lack of a comprehensive examination of the effects that

political discussion has across diverse polities. Second, in my analysis I test directions of the

effects between interpersonal communication and political engagement. This allows me to answer

the question whether the oft-documented relationship between political communication and

engagement is due to the fact that communication enhances engagement or the other way round.

The vast majority of previous research could not answer such questions, mostly because of the

lack of appropriate data. The third contribution of this study lies in providing an explanation of

the mechanism that links interpersonal communication and political engagement. There is limited

reflection on these issues beyond the finding of a significant relationship between the two.

Building on social psychological theories, especially theories of social identity, uses and

gratification, I suggest that political discussion influences forms of political participation and

cognitive involvement with politics through an increase in political interest. Finally, my study

explores the supply side of these politically relevant features of the interpersonal communication

and test the hypothesis that length of democratic tradition predicts them. In the same context, it

investigates generational effects on social networks usage and political discussion patterns

between people who were socialized under authoritarian regimes and those who grew up in

democracies.

To explore these questions I draw on several comparative studies, namely the

International Social Survey Program (ISSP), the Comparative National Election Project (CNEP),

and Citizenship, Involvement, Democracy (CID), and panel data collected in national studies

5

conducted in Japan and Hungary. The results of my investigation show that interpersonal

communication has both a direct and an indirect effect on an individual‟s level of political

engagement. Moreover, these effects are moderated by macro level factors, such as length of

democratic experience, level of economic development, and norms of social and political

communication in the countries analyzed. Specifically, political similarity between respondents

and their intimate, close peers functions as an incentive for electoral participation. This effect

appears to be stronger for those people living in countries with a more recent democratic

experience. Additionally, political similarity with close discussion partners has a positive effect

on political knowledge for those living in older democracies, more economically developed

countries, and societies that have higher averages of political agreement. On the contrary, more

frequent talk with people who are part of generic social networks has a positive effect on political

knowledge for those people who live in less economically developed countries. The indirect

effect of interpersonal communication operates through an increase in levels of political interest

and media attentiveness. More frequent political talk in either intimate or generic settings leads to

an increase in levels of political interest and media attentiveness in all countries analyzed. Both

political interest and media attentiveness are among the strongest antecedents of political

knowledge and participation. However, political agreement in intimate settings appears to

discourage media consumption and this is even more the case in older democracies and more

economically developed countries. Also, political conversations in intimate settings cease to have

a stimulating effect on media attentiveness in those societies characterized by extensive informal

political talk.

When the assumption that interpersonal communication affects political engagement is

questioned, results show mixed evidence. The classic assumption that political talk promotes

knowledge and diversity of political opinions does not hold in the political context of Japan. On

6

the contrary, in my analysis I find that those people who are more politically knowledgeable and

hold a diversity of political opinions are more likely to engage in political conversations with

their peers. For the relationship between political discussion and participation though, results

show that the classic assumption holds when tested in the context of Hungary. Those people who

have more political discussants among their close others participate more often in civic and

political activities. With regard to all these analyses, there is evidence that political discussion

affects both political participation and knowledge through an increase in levels of interest for

politics. The amount of political discussion in one‟s micro social settings signals the importance

attributed to this topic in those social environments. Theories of social identity and those of uses

and gratification would predict that this acts as an incentive for group members to seek political

information either through media or political conversations. My analysis also confirms the

existence of reciprocal effects between engagement in political discussion and levels of political

interest.

Finally, there is evidence that the supply of these politically relevant features of the micro

social settings varies across countries. People who live in countries with a non-democratic past

are embedded in smaller and less diverse social networks, tend to discuss less politics in their

everyday social settings and participate in more politically biased conversations. Within countries

with a more recent democratic experience, people who were socialized during democratic years

are less likely to discuss politics compared to people who were socialized under authoritarian

regimes. This suggests that past legacies might be transmitted between generations, although this

finding might be also indicative of differences in regime type and openness of the political

competition in the aftermath of democratization.

My dissertation proceeds as follows. In the first chapter I introduce the main concepts of

my research and discuss terminological distinctions with regard to their use across existing

7

studies. I give an overview of the conclusions reached in previous research on the social

underpinnings of political engagement and highlight the gaps that I see in these studies, mostly

with regard to their inconsistencies and limitations. I introduce my research questions and give a

brief account of the research design employed in my dissertation. More detailed accounts of

specific hypotheses, data and research design are given in each of the chapters. In chapter two I

present the results of a comparative analysis of the effects that casual political conversation in

close, intimate social settings has on individual electoral participation and political knowledge.

Chapter three complements this analysis with an examination of the effects that political

discussion in generic social settings has on individual voting and political cognition. These

analyses employ data collected in countries with dissimilar histories and political cultures,

including different democratic experiences. In both chapter two and three, I test the moderating

role of macro contextual variables on the relationship between informal political discussion and

engagement. These macro level variables are the length of democratic experience in the countries

analyzed, their level of economic development, and norms of social and political interactions. In

chapter four, the classic direction of influence between political communication and engagement

is examined empirically. Specifically, in this chapter I test for reciprocal effects between political

communication and two forms of cognitive involvement with politics, namely political

knowledge and diversity of political opinions. Chapter five continues the examination of direction

of effects between political communication and another form of individual political engagement,

namely political participation. Chapter six explores the determinants of those politically relevant

features of the interpersonal communication. Specifically, it looks at differences in patterns of

social networks and informal political discussion between old and new democracies. It also test

whether, in countries with a more recent democratic experience, there are significant differences

8

between people who were socialized under authoritarian regimes and those who came of age

under democracies, with regard to social network usage and political discussion patterns.

Throughout my dissertation I employ statistical methods that were previously used by

scholars engaged in research on similar issues. On the one hand, this makes possible a

comparison between my results and the findings of extant research. On the other hand, as

discussed in my chapters, other methods might be more appropriate for the study of the problems

raised by my investigation. Moreover, as highlighted in the last part of each individual chapter,

there are limitations imposed by the use of specific methods.

9

Chapter 1: Social networks, political discussion and engagement: the

state of the art

In this chapter I introduce the main concepts that I employ in my research and highlight

various terminological uses that they received in the previous literature. I discuss the findings of

the extant literature on the sociological underpinnings of individual political engagement and

point at some inconsistencies and gaps in their conclusions. I then present the research design of

my study: research questions, general expectations on the role of interpersonal communication on

individual political engagement, the data and the methods that I employ to explore these issues.

1.1. Main concepts

Social networks

Social networks refer to the social interactions that people have in their everyday lives.

They include, but are not necessarily limited to, interactions that occur within families, groups of

friends, workmates and various voluntary organizations to which individuals belong. Social

networks are seen as sources of social capital (Coleman 1988, Putnam 1995). They provide

resources, such as trust and cooperation, to people who are engaged in regular social interactions

(Coleman 1988, Putnam 1995). Such resources make possible „the achievement of certain ends

that in its absence would not be possible‟ (Coleman 1988: 98). Equally important, social

networks function as channels of information diffusion and, therefore, can contribute to an

increase in levels of political information, knowledge and participation (Granovetter 1973,

Putnam 1995).

10

Recent advancements in network analysis made possible sophisticated research on

diffusion of information, norms and behaviors in large networks (Christakis and Fowler 2009). A

major drawback of such studies, though, is that they are conducted in restricted contexts and,

therefore, their conclusions cannot be generalized. Surveys are the most common method of

collecting social networks data. This way, researchers can gather information on a large number

of individuals and have the ability to generalize their conclusions to the population sampled. On

the other hand, network data collected in surveys gives a partial representation of the micro social

contexts in which individuals reside. This is due to the fact that usually there are limits in the

number of people that can be indicated as part of one‟s networks and very often there is no

information about the relationships within the network of an individual respondent (ego-centric

network).

Some scholars who studied the political role of social networks distinguished between

micro-social contexts, which are those environments in which people reside, such as

neighborhoods, for instance, and social networks, which are those social interactions selected

from these social contexts. This distinction emphasizes that while social contexts are by and large

given – although, they are not totally exogeneous to individual choices – social networks include

those people who are chosen from the inhabitants of these given micro social contexts (Huckfeldt

and Sprague 1987). Granted, in reality, these two concepts overlap and scholars discuss them

separately solely for analytical purposes.

In my research, social networks are considered to be the result of an individual‟s choice,

which is structurally constrained by the more generic social settings that she inhabits. In most of

the previous research, larger social contexts were measured as small territorial units, such as

neighborhoods, for instance. In my research, the macro social units that I investigate are

11

countries, in which people are expected to share similar norms of social interactions and political

communication.

An even subtler distinction is the one between social groups and networks (Huckfeldt et

al. 2004). This distinction is important for understanding another key element of interpersonal

communication that I am discussing below, namely political heterogeneity or cross-pressure. As

Huckfeldt and his colleagues show, there are important, though not so obvious implications of

conceiving individuals‟ interdependence in terms of social networks rather than self-contained

groups (Huckfeldt et al. 2004: 17-8). The understanding of social influences stemming from

individuals‟ embededdness in small groups, in which people are connected among them, is an

understatement of the more fluid social reality. In most of the studies conducted before the

advancements in network analysis, individuals‟ interdependence was discussed in terms of

embeddedness in such groups. The studies conducted by Lazarsfeld and his colleagues

(Lazarsfeld et al. 1944, Berelson et al. 1954), for instance, followed this approach. The same

method was employed in social psychology research (Festinger 1954). The major drawback of

analyzing social influences in politics in the framework of embeddedness in small, self-contained

groups is that it underestimates the more fluid nature of these influences and overlooks the social

interactions that occur outside the boundaries of these pre-defined groups. Therefore, the concept

of „social networks‟ might offer a more accurate description of an individual‟s social interactions.

It also captures differences in the social interactions patterns of individuals who are located in

similar social groups. People who belong to the same social groups might be exposed to different

social influences inasmuch as they attend to social interactions that are not shared by their group

fellows. The social interactions occurring outside the boundaries of the pre-defined groups might

have stronger influences, as emphasized by Granovetter‟s model of „weak ties‟ and Burt‟s

concepts of „bridges‟ and „holes‟ in the social networks. Moreover, influences can occur even in

12

the absence of direct contact and this is captured in Burt‟s model of social equivalence (Burt

1987).

With regard to social networks‟ influence in politics, there is evidence that they provide

mobilization for electoral participation (McCarthy and Zald 1977, McCarthy 1996, Rosenstone

and Hansen 1993, Verba et al. 1995). Social networks work as „micromobilisation contexts‟ for

political engagement (Snow et al. 1980, McAdam 1988, McAdam and Paulsen 1993).

Notwithstanding explicit calls for recruitment that might come through these interactions, some

structural features of the social networks were found to affect individual‟s political behavior and

cognition. They are the size, diversity, strength and density of the social networks.

Individuals can have smaller or more extensive social interactions; these interactions can

provide a protective „cocoon‟ or can expose people to diverse opinions and views. Small social

networks usually contain close, intimate relationships; this coincides with Granovetter‟s concept

of „strong ties‟. Larger, more generic social networks, on the other hand, might include „weak

ties‟ along with the strong ones (Granovetter 1973). Each type of ties serves different purposes.

„Strong ties‟ generate trust and norms of reciprocity, whereas „weak ties‟ connects individuals

with diverse social contexts and give them access to various and potentially non-redundant

information sources (Granovetter 1973). Similarly, Burt discusses about the role of „bridges‟ and

„holes‟ in the social networks, which connects individuals located in different social networks. In

Putnam‟s research, strength of ties enters in his distinction between „bonding‟ and „bridging

social capital‟ and networks. Network density is a concept that social science scholars borrowed

from network analysis literature. It measures how connected network members are. In a dense

network, two individuals share the same associates (Granovetter 1973, Burt 1987), whereas in

sparser networks there are a small number of ties relative to the total number of possible ones.

Network density is important for exposure to political disagreement; in a highly connected

13

network, the chances of exposure to a diversity of political views are severely curtailed. On the

other hand, tight networks have their advantages; they enhance social cohesion and are important

sources of trust and close cooperation. They provide social and political identity and,

consequently, reinforce political views and fuel political engagement.

However, what makes these social interactions particularly prone to create political

influence is the explicit exchange of political opinions and arguments that go on within these

social networks. Some authors discussed about a politically relevant social capital, which does

not come about in the absence of specific political interactions occurring in these networks (Lake

and Huckfeldt 1998).

Political talk

Previous studies employed various terminologies to describe political conversations that

occur among peers as a by-product of their everyday social interactions. Political talk, civic talk,

political discussion, political conversation or informal deliberation, are just a few of them. What

they have in common is that these conversations are neither bound by formal rules nor aimed at

reaching a consensual result. In fact, this is what some authors believe it separates informal

political conversations from other forms of deliberative practices, such as deliberative polls and

citizen juries (Klofstad 2011). As Mutz suggests, the standards of the ideal deliberative settings

are rather difficult to meet in real life (Mutz 2006).

Information about casual political talk is usually collected in surveys through two types of

questions (name-generators) that prompt different relationships. One of them asks respondents to

indicate people with whom they usually discuss politics. This explicit request to nominate

political discussants is more likely to elicit information about respondents‟ „weak ties‟. The other

type of question requests information about respondents‟ close, intimate social interactions and

14

habits of discussing politics with these people. This type of question is more likely to elicit

information about respondents‟ „strong ties‟. However, there is empirical evidence that, in the

American context, the two types of questions generate similar types of relationships (Huckfekdt

and Sprague 1995). People tend to discuss politics with those with whom they do other activities

together. In other words, politics does not lead to a compartmentalization of these relationships.

This can be justified from a rational point of view; specialization of relationships comes with a

cost and, therefore, people are better off if they use the same contacts to serve diverse interests,

such as entertainment and political discussion, for instance (Huckfeldt and Sprague 1995: 17).

A different approach on the role of political discussion is pursued by Eliasoph (1998) and

Walsh (2004), who observed small communities, over a long period of time. Although this

approach produces wealthier information than survey studies do, it has a limited capacity of

generalizing the conclusions beyond the contexts in which data was collected.

The features of political talk that were found to have consequences for individuals‟

political behavior and cognition are the size of these political networks, the frequency of these

political conversations, the level of political expertise and agreement in these networks of

political discussion.

Political homogeneity/heterogeneity

An important element of informal political conversations is whether they expose people to

similar or divergent political views. Whereas the concept of network homogeneity has been

consistently used in the previous literature, the one pertaining to the presence of opposing

political views is marred by both terminological and measurement issues. A look at previous

literature on the social underpinnings of political behavior indicates that the following terms have

15

been used to describe micro social contexts that expose people to views different from their own

ones: cross-cutting social networks (Mutz 2002b), heterogeneous (Huckfeldt and Sprague 1995,

Scheufele et al. 2004, McLeod et al. 1999), ambivalent (Nir 2005), diverse (Huckfeldt and

Sprague 1995) and dangerous (Eveland and Shah 2003, Eveland and Hiveley 2009, Eveland et al.

2010) micro social settings.

In the seminal studies on the role of social contexts in politics, Lazarsfeld and his

colleagues introduced the concept of „cross-pressure‟, understood as an ideological imbalance

created by an individual‟s membership in diverse social groups (Lazarsfeld et al. 1944, Berelson

et al. 1954). Social categories were employed until relatively recently as proxies for individuals‟

ideologies and political preferences. More recently, though, survey studies provide direct

information about exposure to divergent political views and political cross-pressure in

individuals‟ social networks. These studies include measures of respondents' self-declared

partisanship or political leaning and their reports on their relationships' partisan preferences. Such

measures can be used to compute direct measures of political heterogeneity or cross-pressure.

Some authors continued to employ a combination of the two in order to capture both social and

political sources of heterogeneity in people‟s social surroundings (Scheufele et al. 2004, Kwak et

al. 2005).

However, as Eveland and his colleagues suggest, „one of the major limitations of this

survey-based research is that the definition of disagreement is inconsistent across studies […].‟

(Eveland et al. 2010) This concept was measured in the previous literature with reference to

presidential candidate vote choice, partisan preferences and ideological leaning, and socio-

demographics, such as gender or race (Eveland et al. 2010). Moreover, the indicators of „cross-

pressure‟ or heterogeneity used in the previous literature do not distinguish between social

networks that expose individuals to both similar and divergent political views and those in which

16

they experience only political disagreement. Disagreement and cross-pressure was measured

either as number/ proportion of network members holding political views different from

respondents' (Mutz 2002b) or as number/ proportion of network members who support both

similar and different views as compared to respondents (Nir 2005).

Differences in how political disagreement is measured across various studies are likely to

have consequences for the conclusions reached (Eveland et al. 2010). To solve these

terminological confusions, Eveland and Hively (2009) propose a distinction among three types of

political discussion networks, namely 'dangerous', 'safe', and 'diverse'. In all three cases the

reference point in assessing similarity/dissimilarity is respondent‟s partisan position. 'Dangerous

discussion' occurs when network members‟ positions 'conflict with the views or characteristics of

the ego' (Eveland and Hively 2009: 208). This concept is similar to what most of the previous

research has labeled as heterogeneous or „cross-cutting‟ discussions (Leighley 1990, Mutz 2002b,

Scheufele et al. 2004). 'Discussion diversity' is measured as 'the degree to which discussions (or

discussion partners) in various researcher-designated categories are evenly distributed across

those categories' (Eveland and Hively, 2009: 208). Nir's concept of „ambivalence‟ is similar to

discussion diversity (Nir 2005). Finally, 'safe discussion' is the logical „inverse of dangerous

discussion' and includes 'those discussions that coincide with the views or characteristics of the

ego' (Eveland and Hively, 2009: 208). 'Dangerous' and 'safe discussion' are not exclusive

circumstances as involvement in one type does not prevent exposure to the other.

1.2. State of the art

This section presents a review of the studies that investigated the relationship between

features of interpersonal communication in the micro social settings in which people reside and

17

their political cognition and participation. This presentation highlights the main inconsistencies in

the conclusions reached by these studies and some of their limitations.

The understanding of individual political behavior as a socially rooted phenomenon had

its heyday in the 1940s with the research of Lazarsfeld and his colleagues (Lazarsfeld et al.

1944). Despite convincing arguments in favor of considering micro social contexts as important

influences for individual electoral behavior, the social model of politics was soon abandoned and

replaced by models where political behavior was understood fundamentally (and even

exclusively) as an individual act. The last two decades have witnessed an accumulation of

research where individual political behavior and knowledge have been approached from a social

perspective. This revival came from different research areas, such as studies on the role of

personal social networks in politics (Huckfeldt and Sprague 1995), social capital (Putnam 2000),

and deliberative theories of democracy (Fishkin 1991).

This revived interest in the social underpinnings of politics resulted in a wide agreement

in the scholarly literature with regard to the relevance of micro social contexts for normatively

desirable outcomes such as political participation (Lazarsfeld et al. 1944, Berelson et al. 1954,

Huckfeldt and Sprague 1995, Lake and Huckfeldt 1998, Mutz 2002b, 2006, McClurg 2003,

2006a, Huckfeldt et al. 2004), and political knowledge and sophistication (Kenny 1998, Holbert

et al. 2002, Bennett et al. 2000, Eveland 2004, Eveland and Thomson 2006, Eveland and Hively

2009).

This accumulation of evidence on the social nature of political behavior and cognition

revealed potential theoretical inconsistencies and normative tradeoffs (see especially Mutz 2002b,

2006). For instance, some studies highlighted that the type of social setting that furthers tolerance

and knowledge is one that might depress participation and political involvement (Mutz 2006).

Specifically, while some studies found that exposure to politically conflicting views in micro

18

social settings is negatively associated with various forms of participation, (Lazarsfeld et al.

1944, Berelson et al. 1954, Mutz 2002b, 2006), others showed a positive effect of political

disagreement (Leighley 1990, McLeod et al. 1999, Scheufele et al. 2003, Scheufele et al. 2004,

Kwak et al. 2005) or a non-significant one (Horan 1971, Nir 2005). Several explanations of this

phenomenon were proposed. Some of them attributed the diversity and at times contradictory

nature of these findings to the inconsistent way of conceptualizing and operationalizing political

disagreement across different studies (Eveland and Hively 2009, Eveland et al. 2010). Others

suggested that given that 'the “social context” is a multilayered phenomenon', research should

take into account both micro and macro social embeddedness as sources of political influence

(McClurg 2006a: 350).

Research on the social underpinnings of political behavior originates in the studies of the

scholars who became known as the Columbia school of sociologists (Lazarsfeld et al. 1944,

Berelson et al. 1954). Their research focused on the role of interpersonal communication in

electoral campaign. The results of their empirical investigation emphasized the substantial role

played by primary groups in molding partisanships and promoting participation. Specifically,

their studies showed that frequent and intimate interactions with like-minded others contribute to

the maintenance of stable political preferences and voting patterns.

This sociological model of voting established a line of research where direct, frequent and

intimate contact was seen as the main force driving the mechanism of political influence at the

level of primary groups. This model of influence is referred to by Burt (1987) as the social

cohesion model (Coleman 1966). Confronted with the necessity to make a decision, people turn

to their close others and solve the problem through debate at the network level until reaching

agreement. Thus, very often, the result is similarity of attitudes, beliefs and behavior among

network members.

19

Despite its wide acceptance, the social cohesion model has not remained unchallenged.

Burt formulated the structural equivalence model as an alternative framework for explaining

social influence. The core of this model is the similarity of positions occupied in the social

structure. Two people are structurally equivalent if they are similarly connected to the same alters

or if they display „the same patterns of relations with occupants of other positions‟ (Burt 1987:

1291). Structural equivalence is not predicated upon direct contact, but awareness of similarly

located others. The influence is not the result of debate among close others but of the relative

deprivation within status/position experienced by people who acknowledge the existence of

structurally equivalent others and thus their substitutability.

Whereas the rationale of the structural equivalence model is clear and also directly proved

in the case analyzed by Burt (drug adoption among physicians was not the result of discussion

with trusted others but of the feeling of relative deprivation within one‟s own status) it is less

comprehensible when this framework is applied to research on voting behavior (most notably by

Huckfeldt and Sprague 1991, 1995). Huckfeldt and Sprague claim that „a dyad is marked by

structural equivalence to the extent that individuals share the same social relationships‟ (1991:

124). However, as the authors admit, a direct test of the structural equivalence model is quite

impossible due to the lack of appropriate data. Therefore, the model can be approximated by

finding conditions that might give raise to structurally equivalent people. For them, people are

„more likely to be structurally equivalent if they live in the same neighborhood, drink at the same

bars, worship at the same churches, work at the same place of employment, and so on‟ (1991:

125). The rationale implied here is that people embedded in similar contexts would look at each

other‟s behavior based on the assumption of shared common interests. As the authors point out,

20

the crucial difference between the two models is that in the structural equivalence model

influence might be „divorced‟ from strong ties, direct contact, and debate among people.

On the other hand, the studies conducted by Lazarsfeld and his colleagues indicated that

exposure to cross-pressure in the micro social groups to which individuals belong led to delay in

the formation of voting preferences in the electoral campaign and ultimately depressed electoral

participation (Lazarsfeld et al. 1944, Berelson et al. 1954). Unstable voters and defectors were

shown to come mainly from the group of those that either could not recall any political discussion

in their groups or were exposed to conflicting political views (Lazarsfeld et al. 1944, Berelson et

al. 1954). For Lazarsfeld and his colleagues, the source of cross-pressure was the affiliation to

multiple social groups that embodied different and conflicting political views and values. It is

important to keep in mind though, that the conflicting potential of these social allegiances was

entirely assumed by researchers but was impossible to be established based on the available data.

Further research found little if any support for the theory of cross pressure in its original

form or directly measured as exposure to conflicting political views. A replication of the analysis

conducted by Lazarsfeld and his collaborators using similar data sets found no support for the

cross pressure thesis (Horan 1971). His conclusion is that previous results were due to

interpreting social positions as sources of political conflicts. More recent studies included direct

measures of exposure to conflicting political views. They are based on respondents‟ reports on

their political preferences and their perceptions on their discussants‟ political views. However,

the results of these studies are inconsistent. People embedded in politically heterogeneous social

networks measured as amount of conflicting political views appear to be more likely to refrain

from political activity, mainly to avoid putting their relationship at risk (Mutz 2002b: 851, 2006).

In studies where cross-pressure is operationalized as a balance of pros and cons, politically

21

ambivalent settings appear to have no effect on participation, vote choice and crystallization of

vote decision (Nir 2005). Other studies indicate that there is a positive relationship between

exposures to opposing political views (Leighley 1990, McLeod et al. 1999, Scheufele at al. 2004)

and political participation.

Eveland and Hively suggest that this is due to differences in measuring political

disagreement and cross-pressure. As exposed in the previous section of this chapter, they propose

a tripartite reference to these concepts, which include all terminologies that were previously

employed (Eveland and Hively 2009). These are „safe‟, „dangerous‟, and „diverse‟ political

discussion or political networks (Eveland and Hively 2009). However, the root of the inconsistent

results reached by previous studies seems to be deeper than these terminological differences.

Even after harmonizing the terms, some of the noted differences and inconsistencies remain.

„Diverse discussion‟ is found to have a negative effect on political participation in some studies

(Eveland and Hively 2009), while being non-consequential in others (Nir 2005). Moreover,

„dangerous discussion‟ has negative effects on participation in some studies (Mutz 2006) and no

effects in others (Eveland and Hiveley 2009).

The persistence of inconsistencies suggests that some further specifications might be

needed within the concept of micro social contexts. McClurg, for instance, suggested that social

context is a „multilayered phenomenon' (McClurg 2006a: 350) and therefore researchers should

examine sources of influences located at different levels of analysis. Some scholars, for example,

considered the moderating role of macro social contexts, such as social and partisan composition

of the communities where individuals reside, i.e. neighborhoods (Berelson et al. 1954, Huckfeldt

and Sprague 1995, McClurg 2006b) or the distribution of political preferences at the country level

(Anderson and Paskeviciute 2005).

22

In a larger perspective, the suggestion that social context is a „multilayered phenomenon‟

generated further thoughts, especially among comparative scholars, who contemplate the idea that

the effect of interpersonal contexts might be moderated by macro level features, such as social

structures and political cultures. Specifically, Mutz's finding that political disagreement at

network level discourages individual political participation mainly for reasons of social conflict

avoidance elicits a broader reflection on the role of social norms governing debate across various

polities (Mutz 2002b). Peng and Nisbett suggested that exposure to political disagreement might

put a stronger strain on Americans compared to people from other countries (Peng and Nisbett

1999 cited in Mutz 2002b). However, there is no empirical evidence with regard to „the extent to

which political disagreement is deemed socially acceptable.' (Mutz 2002b: 851). One step further,

some scholars have specifically inquired into the role of cultural differences in moderating the

relationship between networks diversity and political tolerance in Japan as compared to the US,

where the bulk of such studies has been previously conducted (Ikeda and Richey 2009).

However, no systematic examination of the moderating role of structural and cultural

differences has been conducted so far. The vast majority of existing studies concerned with the

relationship between social embededdness and individual participation and knowledge were

conducted in the US (but see more recently a few in Japan, Germany, Spain, and the UK). We

thus have sufficient, albeit at times contradictory, evidence that social networks play significant

informational and mobilization roles in a number of advanced democracies. Nevertheless, we

cannot but speculate about the similarity of their roles in transmitting participatory norms and

informational cues in countries that have a more recent democratic experience or a different

socio-cultural makeup. A nomothetic understanding of the social influence in politics is thus

missing and leaves us wondering whether it is always the case that more political talk is linked to

more political knowledge and participation and whether political disagreement provoke more

23

reflection or not across societies that differ in their norms of social interactions and political

cultures.

Cultural studies do specifically emphasize that the way individuals understand,

communicate, and behave depend both on their individual predispositions and values and the

culture to which they belong (Hofstede 2001). In this view, taking political cues from peers, for

instance, may turn out to be a more widespread and pervasive feature of societies that, for

different historical and cultural reasons, place higher importance on interpersonal communication

than on formalized means of communication via media, political parties, and social and civic

organizations. However, a higher appetite for political discussion in a society is not necessarily an

indicator of its efficiency. More debate may not automatically contribute to an increase in

participation and knowledge but may bring about skepticism and demobilization. It is therefore

important that micro social influences are investigated both independent of and in conjunction

with macro level variables susceptible at moderating their political influences.

Previous literature suggested that it maters a lot whether and to what extent people

encounter divergent political views in their everyday interactions (Huckfeldt and Sprague 1995,

Huckfeldt et al. 2002, Huckfeldt et al. 2004, Mutz 2002a, 2006). However, homogeneous

networks appear to be the norm of social communication as people prefer the company of like-

minded others and do their best in preserving the agreeable character of discussion through

mechanisms such as selective exposure, avoidance of disagreement and, ultimately, shifts in their

preferences and views (Heider 1958, Festinger, 1957). Similar propositions seem to hold true as

regards political discussion networks (Bennett et. al. 2000, Mutz 2002a, 2006). On the other

hand, there are studies claiming that since people cannot have an exclusive control over their

social interactions, political disagreement is a quite frequent occurrence of their everyday

contacts (Huckfeldt et al. 2002, 2004, 2005). This is even more the case for those who are highly

24

interested in politics and therefore less likely to avoid this topic of conversation (Huckfeldt and

Mendez 2006).

In addition to the benefits exerted on participation, everyday political conversation

appears to affect positively individuals‟ ability to understand political issues. When people

engage in political talk they become more aware of their views and their rationale, better

informed on various issues and more knowledgeable as regards the opposing views. Although a

series of studies have found a significant relationship between frequency of engaging in political

discussion and being more knowledgeable in matters regarding politics or public affairs generally

(Kenny 1998, Holbert et al. 2002, Bennet et al. 2000, Eveland, 2004, Eveland and Thomson

2006) only a few were able to make a stronger claim about the relationship between the two

based on panel data analysis: „frequency of discussion and discussion elaboration do appear to be

significant causes of political knowledge‟ (Eveland and Thomson 2006: 523). As in the case of

the link between political talk and participation, the study of the relationship between discussion

and knowledge (or information) is fraught with problems of endogeneity, difficult to sort in cross-

sectional analysis.

When partisan composition of political discussion networks is added to the analysis

results indicate that homogenous networks are less likely to promote political knowledge (Mutz

2002b, 2006). This seems to be due to the fact that the company of like-minded peers reinforces

people‟s views and give them fewer opportunities to encounter opposing arguments and learn

new facts. Informal political conversation might bring cognitive benefits others than factual

political knowledge or information. Political talk can spread informational cues or shortcuts that

might operate as functional equivalents of the full information and assist people in the process of

decision making. Scholars of „low rationality‟ in politics have claimed that the limited

information people possess in the political sphere does not incapacitate them in making

25

reasonable decisions (Bartels 1996, Cutler 2002, Lau and Redlawsk 1997, Lupia 1994, Lupia and

McCubbins 1998, Popkin 1991). Their claim is that the use of informational cues helps people to

„vote correctly‟, i.e. make decisions similar to the ones they would have made if they were fully

informed. Even if political discussion networks are seen as potential providers of such

informational cues, little research has been done to single them out and investigate their effect

among other heuristics people use in the process of decision making. Toka‟s findings, though,

revealed that the informational cues provided by casual political conversations are not a substitute

of political knowledge in the process of political decision making at individual level (Toka 2010).

He showed that participation in political discussion promotes political knowledge but there are no

additional benefits of political talk in producing better informed political choices (Toka 2010).

To sum up, I identify three types of gaps in the previous research on social influences in

politics. First, there is a lack of examining cross-national variations in the role played by micro

social contexts on individual political engagement. Most of the previous research was conducted

in a limited number of countries and their results showed inconsistencies on the effects of micro

social settings on individual political behavior and cognition. This led to some reflection on the

moderating role that macro level features, such as social structures and political cultures, might

have on the relationships between micro social contexts and individual political engagement.

However, the lack of appropriate cross-national data precluded a more general examination of the

patterns of micro social influences in politics. Second, most of the previous research was based

on the theoretical assumption that political discussion influences political participation and

cognition rather than the other way round. However, many studies admitted that the reverse

direction of effects is equally plausible. Finally, the third gap that I identify is related to the

limited examination of differences in social networks usage and political discussion patterns

between people living in old and new democracies and, within new democracies, between those

26

people who were socialized under authoritarian regimes and those who came of age in democratic

periods.

1.3. Research questions, design and data

In my dissertation I aim at filling in the identified gaps of the previous literature on micro

social influences in politics. The research questions that I address in this study are the following.

First, are there any general influences that interpersonal communication has on an individual‟s

level of political engagement? Second, do macro level variables, such as length of democratic

experience, level of economic development and norms of social and political communication,

moderate the effects that interpersonal communication have on political engagement? Third, are

there unidirectional or reciprocal effects between features of social and political interaction on the

one side, and political engagement on the other? Fourth, what is the mechanism through which

interpersonal communication contribute to an increase in levels of political engagement? Finally,

what are the determinants of individuals‟ engagement in political conversations and how do they

vary across countries that differ in their democratic traditions?

To answer these questions I draw on data from three comparative studies, namely the

International Social Survey Program (ISSP), the Comparative National Election Project (CNEP),

and Citizenship, Involvement, Democracy (CID), and panel data collected in national studies

conducted in Japan and Hungary. The Social Network Module of the ISSP study was conducted

in two waves, one in 1986 and the other in 2001. The 1986 wave includes seven countries,

namely Australia, the Federal Republic of Germany, Austria, the United States, Great Britain,

Hungary, and Italy. The other wave of the ISSP survey was conducted in 2001 and included

twenty-seven countries, with separate samples for East and West Germany, Great Britain and

27

Northern Ireland. In my analysis I retain the group of European countries that are relevant for my

research questions and the samples from the US and Canada, as representatives of consolidated

democracies that are often discussed together with West European countries.

The CNEP study includes twenty six national election surveys and was conducted in

eighteen countries since 1990. It was conducted in three waves and all of them were concerned

with intermediation processes through which citizens receive information during the electoral

campaign. The selection of the countries does not follow a random procedure and thus the result

is a convenience sample rather than one that is representative for the whole set of extant societies.

However, the countries are quite diverse with regard to their histories, institutions, and political

cultures and this makes the study a relevant source for testing the general character of social

influence in politics. The study includes countries from South America, Southern, Eastern and

Western Europe, East Asia, and the US. It features both newly and recently democratized

countries (Spain, Portugal, Chile, Uruguay, Greece, Bulgaria, and Hungary) and stable

democracies (Italy and the US). This study includes a wide range of countries that vary in their

histories, cultures and institutions. While a few of them have a long democratic history most of

them have a more recent democratic tradition or are not yet fully democratized.

The CID project was conducted between 1999 and 2002 in twelve countries, with separate

samples collected for East and West Germany. Out of the other eleven countries there are six with

a more recent democratic experience (Moldova, Romania, Slovenia, Russia, Spain and Portugal)

and five with a longer democratic past (Denmark, Netherlands, Norway, Sweden, and

Switzerland).

There are various statistical methods that I employ in my dissertation. They are described

in details in each of the chapters. In chapter two and three, the comparative study of micro and

macro level determinants of individual political participation and cognition employs hierarchical

28

methods of analysis. The next two chapters – four and five – examine directions of effects

between political discussion and political participation and cognition, using structural equation

models. Finally, in chapter six I use various types of regression models to test differentials in

countries predicting social network usage and political discussion patterns.

29

Chapter 2: Micro social embeddedness, electoral behavior and

political knowledge: a comparative analysis

In this chapter and the next one I examine the relationships between features of political

discussion that occurs in the micro social settings in which individuals reside and their level of

electoral participation and political knowledge. There are two types of micro social interactions

that I explore. One includes those close, intimate relationships that develop among people who

interact often. These social networks partly coincide with what has been referred to by previous

literature as „strong ties‟ (Granovetter 1973) and „bonding‟ social capital (Putnam 2000). The

model of influence believed to operate in these settings is „social cohesion‟ (Coleman 1988). Its

driving forces are intimacy, frequent interaction, similarity and trust. The other types of micro

social settings that I explore are those that expose people to larger, more diverse social

interactions. These social networks can be thought of as „weak ties‟ (Granovetter 1973) and

suppliers of „bridging‟ social capital (Putnam 2000). In this case, influence stems from

individuals‟ exposure to diverse and non-redundant sources of information that might challenge

their political views.

The effects of political conversation with close, intimate peers on an individual‟s decision

to cast a vote and her level of political knowledge are explored in this chapter. In the next

chapter, I present the results of analyzing the influences that political talk in larger, generic social

networks has on voting and political knowledge. These analyses contribute to existing

scholarship on the political relevance of micro social embeddednes in two ways. First, by

separating between the two types of social interactions, it makes possible an assessment of

differential effects that intimate and generic networks have on voting and political cognition.

30

Most of the existing research analyzed either the effects of one type of social interactions or treat

the two types together. This limits our ability to understand whether political communication per

se furthers political participation and cognition or the interpersonal contexts of these

conversations matter, too. The second contribution of my investigation is that it enlarges the

scope of the previous research, by studying relationships between social networks and political

engagement in countries with different political traditions, including diverse experiences with

democracy. This is particularly relevant for understanding to what extent the influence of political

discussion has a general or contextual character. Most of the previous studies were conducted in

the US and a few others consolidated democracies. This limits our ability to understand whether,

for instance, in newly democratized countries, social networks carry similar political influences

as they were found to do in the US.

The analysis draws on cross-country survey data that include societies that differ in their

political traditions, including different democratic experiences. I test whether the relationships

between micro social context on the one side, and individual turnout and political knowledge, on

the other, significantly differ across countries. In addition to this, I analyze whether these

relationships are moderated by macro level factors pertaining to the democratic tradition, level of